1. Field of the Invention
The present invention relates to the field of memory chips.
2. Discussion of Related Art
A known integrated memory IC 100 that is a writeable memory of the DRAM type is shown in FIG. 1. Such a dynamic random access memory (DRAM) chip 100 includes a plurality of memory storage cells 102 in which each cell 102 has a transistor 104 and an intrinsic capacitor 106. As shown in FIGS. 2 and 3, the memory storage cells 102 are arranged in arrays 108, wherein memory storage cells 102 in each array 108 are interconnected to one another via columns of conductors 110 and rows of conductors 112. The transistors 104 are used to charge and discharge the capacitors 106 to certain voltage levels. The capacitors 106 then store the voltages as binary bits, 1 or 0, representative of the voltage levels. The binary 1 is referred to as a “high” and the binary 0 is referred to as a “low.” The voltage value of the information stored in the capacitor 106 of a memory storage cell 102 is called the logic state of the memory storage cell 102.
As shown in FIGS. 1 and 2, the memory chip 100 includes six address input contact pins A0, A1, A2, A3, A4, A5 along its edges that are used for both the row and column addresses of the memory storage cells 102. The row address strobe (RAS) input pin receives a signal RAS that clocks the address present on the DRAM address pins A0 to A5 into the row address latches 114. Similarly, a column address strobe (CAS) input pin receives a signal CAS that clocks the address present on the DRAM address pins A0 to A5 into the column address latches 116. The memory chip 100 has data pin Din that receives data and data pin Dout that sends data out of the memory chip 100. The modes of operation of the memory chip 100, such as Read, Write and Refresh, are well known and so there is no need to discuss them for the purpose of describing the present invention.
A variation of a DRAM chip is shown in FIGS. 5 and 6. In particular, by adding a synchronous interface between the basic core DRAM operation/circuitry of a second generation DRAM and the control coming from off-chip a synchronous dynamic random access memory (SDRAM) chip 200 is formed. The SDRAM chip 200 includes a bank of memory arrays 208 wherein each array 208 includes memory storage cells 210 interconnected to one another via columns and rows of conductors.
As shown in FIGS. 5 and 6, the memory chip 200 includes twelve address input contact pins A0-A11 that are used for both the row and column addresses of the memory storage cells of the bank of memory arrays 208. The row address strobe (RAS) input pin receives a signal RAS that clocks the address present on the DRAM address pins A0 to A11 into the bank of row address latches 214. Similarly, a column address strobe (CAS) input pin receives a signal CAS that clocks the address present on the DRAM address pins A0 to A11 into the bank of column address latches 216. The memory chip 200 has data input/output pins DQ0-15 that receive and send input signals and output signals. The input signals are relayed from the pins DQ0-15 to a data input register 218 and then to a DQM processing component 220 that includes DQM mask logic and write drivers for storing the input data in the bank of memory arrays 208. The output signals are received from a data output register 222 that received the signals from the DQM processing component 220 that includes read data latches for reading the output data out of the bank of memory arrays 208. The modes of operation of the memory chip 200, such as Read, Write and Refresh, are well known and so there is no need to discuss them for the purpose of describing the present invention.
A variation of the SDRAM chip 200 is a double-data-rate SDRAM (DDR SDRAM) chip. The DDR SDRAM chip 300 imparts register commands and operations on the rising edge of the clock signal while allowing data to be transferred on both the rising and falling edges of the clock signal. Differential input clock signals CLK and CLK(bar) are used in the DDR SDRAM. A major benefit of using a DDR SDRAM is that the data transfer rate can be twice the clock frequency because data can be transferred on both the rising and falling edges of the CLK clock input signal.
It is noted that new generations of memory systems that employ SDRAM and DDR SDRAM chips are increasing their frequency range. Currently, SDRAM and DDR SDRAM chips are unable to determine the frequency at which they are operating in a particular memory system. As the frequency range of the memory system widens, it can pose some problems for the SDRAM and DDR SDRAM chips. For example, a DDR SDRAM chip has to time operations between different clocking domains. It is known that the clocking domains change their relative timing to one another as a function of the operating frequency of the memory system. This change in relative timing is illustrated in FIGS. 7 and 8.
In the case of a slow operating frequency, such as e.g 66 MHz, the system clock signal VCLK is directed to the clock pin of the DDR SDRAM. The system clock signal VCLK generates within the DDR SDRAM an internal clock signal ICLK that clocks the central command unit of the DDR SDRAM. This means that all internal commands generated by the central command unit are synchronized with the internal clock signal ICLK. As shown in FIG. 7, while the internal clock signal ICLK has the same frequency as the system clock signal VCLK, it lags the system clock signal VCLK by a constant amount tMAR2. The lag is caused by several gate and propagation delays. This lag results in a phase shift between ICLK and VCLK that grows in magnitude as the frequency of the clock signals is raised. This phase shift increase is a result of the relation of the constant tMAR2 to the cycle time that decreases with an increase in the clock frequency.
As shown in FIG. 7, a second internal clock signal DCLK is generated by a DLL of the DDR SDRAM. The internal clock signal DCLK and the system clock signal VCLK each have the same frequency. However, the internal clock signal DCLK is advanced with respect to the system clock signal VCLK by a constant amount tMAR1 that is dependent on the chip temperature, process variation and the operating frequency. The purpose of advancing the internal clock signal DCLK relative to the system clock signal VCLK is to time internal events within the DDR SDRAM so that they are edge aligned with the system clock signal VCLK when observed at the external DDR SDRAM pin.
As shown in FIG. 7, the signal SIGclk1 is generated synchronous to the clock signal ICLK. Next, the signal SIGclk1 is synchronized with and handled to the internal clock signal DCLK. As shown in FIG. 7, the signal SIGclk2 shows the timing of the signal after latching (synchronizing) the signal SIGclk1 to the internal clock signal DCLK domain. Signal SIG′clk2 shows the signal SIGclk2 after being shifted by one clock cycle DCLK.
As shown in FIG. 8, a different situation occurs when the system operates at a fast operating frequency, such as e.g. 200 MHz. In particular, while the internal clock signal ICLK still has the same frequency as the system clock signal VCLK, it lags the system clock signal VCLK by a constant amount tMAR2 that results in a greater phase delay when compared with the slow frequency case of FIG. 7. In addition, while the internal clock signal DCLK and the system clock signal VCLK each have the same frequency, the internal clock signal DCLK is advanced with respect to the system clock signal VCLK by a constant amount tMAR′1 that results in a greater phase delay when compared with the slow frequency case of FIG. 7. As shown in FIG. 8, the signal SIGclk1 is generated synchronous to the clock signal ICLK. Similarly, this signal SIGclk1 is synchronized and handled to the DCLK. SIGclk2 shows the timing of the signal after latching (synchronizing) it to the DCLK domain. SIG′clk2 shows the signal SIGclk2 after shifting it by one clock cycle of DCLK. The end result is that the relative timing of the clock signals ICLK and DCLK is drastically different when compared with the slow frequency case.
With the above-described disparity in the relative timing it makes it very difficult to run commands within the DDR SDRAM in a consistent manner independent of the operating frequency of the system. For example, suppose that an output signal of the DDR SDRAM needs to be observed three VCLK cycles after the generation of the signal SIGclk1. If the system was in the slow frequency mode, then the output signal would occur upon the DDR SDRAM chip counting the four DCLK pulses T0, T1, T2 and T3. In contrast, the output signal would occur after the chip counted only the three DCLK pulses T1, T2 and T3 in the fast frequency mode. Thus, the DDR SDRAM chip is unable to consistently run the output command based solely on the number of DCLK pulses counted. This limits the maximum operation frequency in which the DDR SDRAM can be operated within a DDR system. In addition, it limits the types of products run by the memory chip. In particular, a memory chip is able to run products that operate within a particular frequency range while the memory chip is unable to run other products that operate outside the particular frequency range.